Wireless distributed intelligence in personal applications

(1)

Wireless Distributed

Intelligence in Personal Applications

ACTA WASAENSIA 393

COMPUTER SCIENCE

(2)

of the Board of the Faculty of Technology of the University of Vaasa, for public dissertation in Anvia Lecture room (F141)

on the 4th of December, 2017, at noon.

Reviewers Dr Smail Menani

Vaasa University of Applied Science

School of Technology/Information Technology Wolffintie 30

FI-65200 V^AASA FINLAND

Dr Ali Hazmi

Wireless and antenna expert at Huawei FINLAND

(3)

Julkaisija Julkaisupäivämäärä

Vaasan yliopisto Marraskuu 2017

Tekijä(t) Julkaisun tyyppi

Heikki Palomäki Väitöskirja

Orcid ID Julkaisusarjan nimi, osan numero Acta Wasaensia, 393

Yhteystiedot ISBN

Vaasan yliopisto Teknillinen tiedekunta Tietotekniikka

PL 700

FI-65101 VAASA

978-952-476-786-6 (painettu)

978-952-476-787-3 (verkkoaineisto) ISSN

0355-2667 (Acta Wasaensia 393, painettu)

2323-9123 (Acta Wasaensia 393, verkkoaineisto)

Sivumäärä Kieli

158 Englanti Julkaisun nimike

Langaton hajautettu älykkyys eri sovelluksissa Tiivistelmä

Tietokoneet ovat historian kuluessa kehittyneet keskustietokoneista hajautettujen, langattomasti toimivien järjestelmien suuntaan.

Elektroniikalla toteutetut automaattiset toiminnot ympärillämme lisääntyvät kiihtyvällä vauhdilla. Tällaiset sovellukset lisääntyvät tulevaisuudessa, mutta siihen soveltuva tekniikka on vielä kehityksen alla ja vaadittavia ominaisuuksia ei aina löydy. Nykyiset lyhyen kantaman langattoman tekniikan standardit ovat tarkoitettu lähinnä teollisuuden ja multimedian käyttöön, siksi ne ovat vain osittain soveltuvia uudenlaisiin ympäristöälykkäisiin käyttötarkoituksiin.

Ympäristöälykkäät sovellukset palvelevat enimmäkseen jokapäiväistä elämäämme, kuten turvallisuutta, kulunvalvontaa ja elämyspalveluita.

Ympäristöälykkäitä ratkaisuja tarvitaan myös hajautetussa automaatiossa ja kohteiden automaattisessa seurannassa.

Tutkimuksen aikana Seinäjoen ammattikorkeakoulussa on tutkittu lyhyen kantaman langatonta tekniikkaa: suunniteltu ja kehitetty pienivirtaisia radionappeja, niitten ohjelmointiympäristöä sekä langattoman verkon synkronointia, tiedonkeruuta ja reititystä. Lisäksi on simuloitu eri reititystapoja, sisäpaikannusta ja kaivinkoneen kalibrointia soveltaen mm. neurolaskentaa. Tekniikkaa on testattu myös käytännön sovelluksissa,

Ympäristöälykkäät sovellusalueet ovat ehkä nopeimmin kasvava lähitulevaisuuden ala tietotekniikassa. Tutkitulla tekniikalla on runsaasti uusia haasteita ihmisten hyvinvointia, terveyttä ja turvallisuutta lisäävissä sovelluksissa, kuten myös teollisuuden uusissa sovelluksissa, esimerkiksi älykkäässä energiansiirtoverkossa.

Asiasanat

Langaton verkko, ympäristöälykkyys, paikannus, reititys, synkronointi, radiokontrolleri, simulointi, tekoäly

(4)

(5)

Publisher Date of publication

Vaasan yliopisto November 2017

Author(s) Type of publication

Heikki Palomäki Doctoral thesis

Orcid ID Name and number of series

Acta Wasaensia, 393 Contact information ISBN

University of Vaasa Faculty of Technology Computer Science P.O. Box 700 FI-65101 Vaasa Finland

978-952-476-786-6 (print) 978-952-476-787-3 (online) ISSN

0355-2667 (Acta Wasaensia 393, print)

2323-9123 (Acta Wasaensia 393, online)

Number of pages Language

158 English Title of publication

Wireless Distributed Intelligence in Personal Applications Abstract

The development of computing is moving from mainframe computers to distributed intelligence with wireless features. The automated functions around us, in the form of small electronic devices, are increasing and the pace is continuously accelerating. The number of these applications will increase in the future, but suitable features needed are lacking and suitable technology development is still ongoing. The existing wireless short-range standards are mostly suitable for use in industry and in multimedia applications, but they are only partly suitable for the new network feature demands of the ambient intelligence applications.

The ambient intelligent applications will serve us in our daily lives:

security, access control and exercise services. Ambient intelligence is also adopted by industry in distributed amorphous automation, in access monitoring and the control of machines and devices.

During this research, at Seinäjoki University of Applied Sciences, we have researched, designed and developed short-range wireless technology:

low-power radio buttons with a programming environment for them as well as synchronization, data collecting and routing features for the wireless network. We have simulated different routing methods, indoor positioning and excavator calibration using for example neurocomputing. In addition, we have tested the technology in practical applications.

The ambient intelligent applications are perhaps the area growing the most in information technology in the future. There will be many new challenges to face to increase welfare, health, security, as well as industrial applications (for example, at factories and in smart grids) in the future.

Keywords

Wireless Network, Ambient Intelligence, Positioning, Routing, Synchronizing, Radio Controller, Simulation, Artificial Intelligence

(6)

(7)

ACKNOWLEDGEMENT

The research on distributed intelligence, related to the aim of this thesis, was started at the end of the 1980’s, when I developed small user programmable controller logic devices communicating in a wired network. At the beginning of the 2000’s, when I started my postgraduate studies, the target of interest was very distributed automation and the possibilities of wireless communication. The electronics laboratories at Seinäjoki University of Applied Sciences with their modern equipment have been a good environment for the research of new technologies. The funding for postgraduate research at the university and the practical courses with students have been the driving force in the research into wireless technology. I am very grateful to my staff and colleagues for this possibility and their understanding about my research work and the new technology.

I am grateful to the advisor Professor Mohammed Elmusrati at Vaasa University.

He has been very interested in my research and helped me to complete this doctoral thesis with guidance and worthwhile feedback. Thanks to reviewers Dr Smail Menani and Dr Ali Hazmi for valuable comments. Thanks to Professor Lauri Sydänheimo at Tampere University of Technology: he guided me to write my licentiate thesis. I also wish to thank Professor Markku Kivikoski, whose guidance and support of postgraduate studies in Seinäjoki have made it possible to continue this research. Thanks to Reino Virrankoski at Vaasa University: he organizes resources to develop and test wireless technology. Thanks also to John Pearce: he was willing to proofread and correct my English language.

The contribution of students Marko Huhta, Matti Tassi and Matti Ventä has also been significant in software development and routing simulations. Numerous other students at Seinäjoki University of Applied Sciences helped me while testing the developed technology

Special thanks are due to my family: my wife and children. They have been patient and understanding when I have spent time in my research work. I give the biggest thanks to my God and Saviour who has given me the reason to live and motives to be interested in new possibilities and technologies.

Seinäjoki, Finland, October 2017 Heikki Palomäki

(8)

(9)

Figures

Figure 1. The major trends in computing (Weiser, 1996) ... 6

Figure 2. Estimation of computing development lines ... 7

Figure 3. Smart dust mote (Hoffman, 2003) ... 11

Figure 4. RFID μ-Chip (Hitachi ltd, 2003) ... 12

Figure 5. BLE channels (Nilsson, 2013) ... 15

Figure 6. ZibGee protocol architecture (Gomez & Paradells, 2010) ... 16

Figure 7. IEEE 802.15.4 channels (MaxStream, 2007) ... 17

Figure 8. 6LoWPAN protocol architecture (Gomez & Paradells, 2010) ... 18

Figure 9. Z-wave protocol architecture (Gomez & Paradells, 2010) ... 19

Figure 10. Z-Wave network example (Zensys A/S, 2006) ... 19

Figure 11. ANT layer alternatives (Dynastream Innovation Inc., 2014) ... 21

Figure 12. ANT application examples (ANT wireless, 2017) ... 22

Figure 13. TUTWSN topology (Hämäläinen & Hännikäinen, 2007) 22 Figure 14. Soft computing concept ... 37

Figure 15. Modelling methods ... 38

Figure 16. Neural network with one hidden layer ... 39

Figure 17. Neural network learning process ... 40

Figure 18. Prediction with neural network ... 40

Figure 19. Exact control principle of non-linear process ... 42

Figure 20. Fuzzy control principle ... 42

Figure 21. Ant Algorithm (Bonabeau;Dorigo;& Theraulaz, 1999, ss. 26-31) ... 44

Figure 22. Solution with standard tools ... 52

Figure 23. Life cycles and investments of technologies ... 53

Figure 24. The scheme of the wireless development project ... 55

Figure 25. Minimized connection schema ... 57

Figure 26. Minimized layout variants ... 57

Figure 27. The optional modules of the teaching kit ... 57

Figure 28. Key chain module and hand-held modules ... 58

Figure 29. USB stick and DIN rail connector ... 58

Figure 30. nRF24LE1 based node connection schema ... 59

Figure 31. nRF24LE1 based SURFnet module platforms ... 59

Figure 32. Modular SURFnet button (Palomäki, 2011a) ... 60

Figure 33. The smallest SURFnet button ... 60

Figure 34. USB-SPI bridges ... 61

Figure 35. Power consumption of a single active period ... 65

Figure 36. Normal case and retransmit case... 66

Figure 37. Communication jam and retransmit delays ... 67

Figure 38. Simulation software ... 69

Figure 39. Flooding of the routing question (Ventä, 2007) ... 70

Figure 40. Broadcasting route query ... 71

Figure 41. Start state of the routing simulation ... 72

Figure 42. After removing nodes 1 and 58 ... 72

Figure 43. After moving the source and one try ... 73

(13)

Figure 44. After moving the destination ... 73

Figure 45. New start state ... 74

Figure 46. After moving every node in the route ... 74

Figure 47. Start setup of simulation ... 78

Figure 48. Passive positioning: neighbours in relation to the area ... 80

Figure 49. Active positioning: neighbours in relation to the centre range ... 81

Figure 50. Push and pull corrections ... 82

Figure 51. Result example when the MSDE = 126 ... 83

Figure 52. Result example when the MSDE = 22 ... 83

Figure 53. Result example when the MSDE = 4.0 ... 84

Figure 54. Result example when the MSDE = 2.6 ... 84

Figure 55. Positioning with range 5 ... 85

Figure 59. Comparison of the methods ... 87

Figure 60. Indoor positioning area ... 89

Figure 61. Anchor gain tuning, case 1 ... 90

Figure 62. Anchor gain tuning, case ... 91

Figure 63. Anchor gain tuning, cases 3 and 4 ... 91

Figure 64. Excavator bucket tip position calculation ... 93

Figure 65. Neurocomputing iteration loop for excavator tuning .. 94

Figure 66. Simulation loop for tuning method ... 96

Figure 67. Case 1 with 4 test points ... 97

Figure 70. Blade vibration monitoring test setup (Virrankoski, 2012) ... 101

Figure 71. Measures accelerations, while no rotation (Palomäki, 2011a) ... 101

Figure 72. Measured accelerations, while rotating (Palomäki, 2011a) ... 101

Figure 73. Greenhouse sensor (Palomäki, 2013, ss. 292-293) ... 102

Figure 74. Sensor results from air (left) and soil (right) (Palomäki, 2011a) ... 103

Figure 75. Graphical and numerical view (Palomäki, 2011a) ... 103

Figure 76. Positioning in the cattle house (Palomäki, 2011a) ... 104

Figure 77. An example situation in cattle house ... 104

Figure 78. Test case structure ... 106

Figure 79. Node layout ... 107

Figure 80. Router with SPI-USB bridge powering ... 107

Figure 81. Sink node with a SPI-USB bridge ... 108

Figure 82. Star topology ... 108

Figure 83. Star topology propagation delays using RC oscillator ... 110

Figure 84. Topology with 3 parallel routers using RC oscillator . 111 Figure 85. Propagation delays with 3 parallel routers using RC oscillator ... 112

(14)

Figure 86. Topology with 2 parallel and 1 serial router using the

RC oscillator ... 113

Figure 87. Propagation delays with 2 parallel and 1 serial router ... 114

Figure 88. Topology with 3 serial routers ... 115

Figure 89. Topology with 3 parallel routers using a RC oscillator ... 116

Figure 90. Propagation delays in star topology with crystal timing ... 117

Figure 91. Excavator model with wireless acceleration sensors . 118 Figure 92. Wireless acceleration sensors and USB-bridge ... 118

Figure 93. User interface of the python program ... 119

Figure 94. Practical test with 4 reference points and max. ±180º ... 119

Figure 95. Practical test with 4 reference points and max. ±15º error ... 120

Figure 96. Practical test with 3 reference points and max. ±30º error ... 120

Figure 97. Investments in traditional automation ... 126

Figure 98. Investments in amorphous automation ... 127

Tables

Table 1. Comparison between RF controllers ... 26

Table 2. Low-power comparisons ... 63

Table 3. Real and estimated data for simulation input ... 96

Table 4. Simulation cases and results ... 99

Table 5. Different modes (Nordic Semiconductor, 2010b, s. 183) ... 105

Table 6. Star topology test results using RC oscillator ... 109

Table 7. Topology with 3 parallel routers using RC oscillator ... 111

Table 8. Topology with 2 parallel and 1 serial router using RC oscillator ... 113

Table 9. Topology with 3 serial routers using RC oscillator ... 115

Table 10. Star topology with crystal timing ... 117

(15)

Abbreviations

Ambient intelligence

Intelligent electronic devices in our ambient and in us

Amorphous computing

Very large number of identical computers or tasks

Amorphous automation

Very distributed automation

Distributed intelligence

Computing is divided into small independent units

Exergame Outdoor exercise game controlled by information technology

Flat mesh Network without any hierarchy

Positioning To find coordinates, where an object is located 6LowPAN Internet over Low power Wireless Personal Area

Networks

API Application interface

ANT Wireless sensor network technology for fitness applications

DD Direct Diffusion

DLL Data Link Layer in communication protocol stack

FFD Full Function Device

FIFO First In First Out

GPS Global Positioning System

ID Identification number

ISM Industrial, Scientific and Medical

LNA Low-Noise Amplifier

LR-WPAN Low-Rate Wireless Personal Area Network

MAC Medium Access layer in communication protocol stack MEMS Micro Electro Mechanical Systems

(16)

MSDE Mean value of Square of Distance Errors

NFC Near Field Communication (mostly RFID technology)

OEM Own Electronic Manufacturer

PHY Physical layer in communication protocol stack PLC Programmable Logic Controller

PLL Phase Locked Loop

RF Radio Frequency

RFD Reduced Function Device

RFID Radio Frequency Identification RSSI Received Signal Strength Indication SCADA Supervisory Control and Data Acquisition

SoC System on Chip

SOM Self-Organized Map

SPI Serial Peripheral Interface

TUTWSN Tampere University of Technology Wireless Sensor Network

UWB Ultra Wide Band

WLAN Wireless Local Area Network

(17)

connection between people and things around us is necessary for living. The basis for the development of our civilisation is to generate and to share information. The required information is more and more in digital format. Information sustains the structure of society as well as our daily lives. It is becoming more and more important how to connect information processing units with other ones and how intelligent these units are. The computing devices are no longer only data transfer and calculation units, but are also able to make independent decisions and be capable of information filtering.

Information transfer makes it possible to monitor and control processes remotely without having a physical presence. In novel IoT applications, the control of traffic lights, bus timetable displays and other public traffic are all realized in smart cities.

The balance between energy producers and users can be controlled remotely in smart grids, especially private energy producers and renewable energy sources. In smart homes domestic appliances, temperatures and ventilation can be controlled via IoT applications without having a physical presence. Summer houses can be safely left unoccupied by remote monitoring their conditions and controlling the lights etc. This all increases security and people’s ease of living.

The topic of this thesis is to show new areas and applications that are possible with the development of: wireless technology, distributed computing and artificial intelligence. Traditional technology is safe and well known. Automation and our whole life seems to be more risk-free by using only traditional technology in industry and in household management. Actually, to stay in traditional technology results, in the long run, in a reduced ability to compete and an increase in risks when managing our changeable world. In addition, one aim of this thesis is to open people’s vision to new possibilities and to prepare for future automation and management styles that lead to successful and safe living in the future.

Money controls the sharpest tip of technology development. It means that the first winners are those who can invest a lot and have the motives to do so. For this reason, new development serves mostly industry, the military and space technologies. These are good technology drivers, but the results of technology reach the daily life of people much later and only in the form of mass production where companies can gather a lot of profit. Related research groups at universities also perform high scientific analysis, but practical applications are either missing or are too complicated and have too high a cost to be used in household

(18)

management. This thesis provides a simple but efficient technology and shows application possibilities to use them in the daily lives of people in personal areas as well as in small companies. It is very important that people are winners in new technology development, not only organizations.

The most straightforward way to become familiar with new technology and electronics is to develop and use the most efficient structures first (in terms of simplicity and cost) and thus to discover new possibilities. After studying the technology, the current state of the art and the new information processing methods, your eyes are then opened to see and imagine new ways to use this knowhow in new demanding situations. This thesis describes the existing technology, the physical electronic development needed to have a testing platform, simulations to test the functionality of the planned methods as well as application ideas. Some methods realized in the study case applications are based on the developed platforms. The purpose of the simulations and the study cases is first to show that interesting theoretical methods can also function in practice. Second, the purpose is to prove that smart applications can be made in a simple, low-cost format that can be useful in the everyday environment.

The programming language is standard C for radio controllers. While a PC is collecting data from a wireless network the programming language is mostly C# (C sharp) and in one case Python. As an output, the practical cases generate data tables, which is shown in a visual form using Excel charts. Excel and C# language are the calculation tools in the simulation cases. Excel macros are also useful to control calculations. Simulations generate tables and are visualized in Excel charts.

The contributions to this thesis were realized over several years while I was principal lecturer in electronics at Seinäjoki University of Applied sciences. The development, research and test work for this thesis started in 2004 when studying the simplest wireless technology. The teaching of embedded systems was strongly focused on electronic development; therefore, the aim was to develop all of the electronics by ourselves and to utilise them also in teaching. All circuit board layouts are my own design, including also the embedded system teaching kit. The focus was not on ready-made electronic modules and communication standards, because we had development resources and motives to make them ourselves. The students made a significant contribution to the work by developing a programming environment to edit software to be used for a radio controller. The research flows parallel with some courses, therefore it was important to develop and to use only free software tools without license limits. The first wireless transceiver circuit board was the optional module for the embedded system teaching kit. Later in 2006 the electronics was developed in a project funded by LifeIT Oyj. The goal was

(19)

to develop wireless game modules for children and this technology was tested for the first time in a real environment. A routing method simulation and some practical tests were carried out in a student’s thesis. This simulation showed some new aspects that need to be recognize in networks with a large number of nodes.

Connected to a bus technology course, a student class tested radio communication protocol in a larger network. It gave a better understanding of communication range in a real environment.

The university offers teaching-free research periods for study and written work. As a result, I had my licentiate thesis ready in 2008 (Palomäki, 2008). To make a better contribution to research work funded projects are needed. Our university had a cooperation project GENSEN with Vaasa University and Aalto University.

The project gave extra resources to develop network protocols. The Vaasa University people organized some study cases to test the developed wireless ideas in practice. It provided the way to get a practical and very useful response to fix radio communication, low-power features and routing methods. These contributions significantly supported my research and development work. The second teaching-free research period made it possible to contribute with additional network tests and thesis writing.

The thesis structure is as follow: Chapter 2 includes the history of distributed computing development and corresponding background and comparisons.

Chapter 3 includes existing wireless standards, ways to use them and routing algorithms. Chapter 4 describes information-processing methods of artificial intelligence including fuzzy logic and neural networks. Chapter 5 estimates the challenges of distributed automation in future applications. Chapter 6 describes the development of electronic modules with layouts and features. Chapter 7 includes all simulations of wireless networks, its usage and applications. Chapter 8 includes some test cases based on simulations combining real networks with smart routing, positioning and measuring methods. In addition, it includes some application possibilities not tested in practice. Chapter 9 is the conclusions and discussion about topics section.

(20)

2 DISTRIBUTED INTELLIGENCE

A very interesting development line is the roles of computing units that generate, store and use information. The first principle in history was to centralize information and decision making capacity. In this case, connections are in a star format: all data generating and exploiting units must connect themselves with a mainframe computer. This structure is simple to keep control of, but it is vulnerable: one single error in the mainframe computer stops the whole information system and the functions depending on it. In many cases, like in the instance of space research, it is desirable that the remote system continues functioning on its own, making its own decisions and connecting data regardless of the failing connection with the command centre. Actually, in many applications the trend is toward distributed intelligence in which information producers and exploiters communicate with each other directly without mainframe data storage.

The communication backbone is today mostly based on the Internet Protocols (IP).

In addition, even smaller units have the feature to connect to the network. The Internet of Things (IoT) is one good example of this development line.

Reliable communication between the system entities is crucial for successful system integration. According to CISCO, during last year (2016), the total IP traffic has surpassed 1.1 zettabyte (billion terabytes). This number will further increase to about 2.3 zettabyte per year by 2020. About 70% of the total IP traffic will be handled in the last mile by wireless transceivers (Elmusrati, 2017).

Wireless communication makes it possible to connect mobile objects, like phones, navigators, control tags and vehicles. In this way, it is simpler to realize distributed computing without wiring restrictions and costs. Generally speaking, wireless communication has several challenges when compared to wired communication.

For example, wireless communication has higher packet losses because of channel fading, it suffers from higher latency because of competing for limited bandwidth resources, and wireless communication is generally less secure. Hence, it is important that the remote system is able to function with local connections and with a limited central connection. One of the most interesting trends is ambient intelligence applications, where all objects around can communicate wirelessly and function intelligently independently. The personal examples of these are access control, health monitoring, modern games, smart clothes etc. A novel trend seems to be managed chaos, in which all objects have some wireless connections but are intelligent enough to function by themselves. Social insects and cells in tissues function in this managed chaos style. Nature is an important example for new technology: as described in the old wisdom: “Go to the ant, you sluggard;

consider her ways and be wise” (Solomon, 1965, s. 705).

(21)

This thesis describes wireless technology features suitable for new ambient intelligent applications: ambient intelligence is embedded and invisible computing and communication functions implemented in our neighbourhood: clothes, tools, domestic appliances, buildings, furniture, traffic signs etc. Ambient intelligence is normally distributed into very small units that need to be serviced seldom or not at all. The aim of this research work is to estimate the demands, possibilities and features of short distance wireless technology.

The wireless technology today is mostly multimedia communication, where the main purpose is to carry voice, video and files via a wireless channel. This kind of wireless technology is not discussed in this thesis. The focus is on a Low-Rate Wireless Personal Area Network (LR-WPAN).

2.1 History and future of computing

The direction history has taken must be known to know possible future directions.

The current technology is mostly based on history and this can be a disadvantage for technology today. The important question is how to develop wireless technology to fit future demands (Palomäki, 2008, ss. 2-4).

2.1.1 From mainframe to interactive life

One good principle for understanding the development of computers is to estimate how large an area a single computing unit covers in the life of people. Weiser (1996) sees major trends in computing as three phases: mainframe, PC and ubiquitous computing (Figure 1).

(22)

Figure 1. The major trends in computing (Weiser, 1996)

Vertegaal (2003) describes the computing trends in 4 steps based on how many user interfaces one single person uses: 1. many uses one, 2. one uses one, 3. one uses many and 4. many use many. I also see the history of computing in four development phases, but based on the user interface instead; in the first phase, mainframe computers need most knowhow for the user interface. In the last phase are devices that are fully embedded and function independently without any user interface (Figure 2).

In the first development line, one single computer covers or fulfils the needs of one university or one factory. This was the development from programmable electromechanical calculators to mainframe computers. At that time, the developers believed that fewer than ten computers would be enough to fulfil all the computing needs of the world. The mainframe computers were very complicated to use and they needed scientists as operators.

(23)

Figure 2. Estimation of computing development lines

The second development line is microprocessor-based technology, starting from the Intel 4004 processor and going on to the personal computers of today. This development line fulfils the need of a single work place or one family. The effective use of a personal computer needs some studying of informatics.

Another microprocessor development line at the same time is automation applications. Multifunctional microcontrollers replace microprocessors to be suitable for small automation. The next development step was single-board computers for universal use in automation and information technology. The most powerful devices in this development line are Scalable programmable logics for industrial automation.

The next development line goes through personal multimedia devices: mobile phones, music and DVD players, tablets and navigators. One person needs one or more devices of this kind. Using a multimedia device is simpler than using a PC. If you can read and write, you can also use the basic features of a multimedia device.

However, you must be able to read and understand the user manual and you must be able to use a complicated keyboard. For babies, demented people and disabled persons the use of multimedia devices can be too difficult.

The last development line is ambient intelligence, which is located around us invisibly. The computing intelligence embeds in keys, locks, domestic equipment, wristwatches, traffic lights, cash desks etc. Future applications include for example the monitoring of children, exercise games, health control and condition monitoring in industry. One single person or work place needs numerous

(24)

embedded computing and communicating devices. This newest development line is still seeking its directions. We expect that in the future ambient intelligence technology will capture new application areas and can soon be the fastest growing area in informatics.

Every technology line has started with simple electronics based on earlier technology and then developed into more powerful devices that perform complex functions. The first line includes some tens of computers. The second line contains some billions of PCs and control computers today. The number of mobile connected devices has been estimated to be 11.6 billion by 2021 (Cisco, 2017). It is perhaps impossible to count the number of ambient intelligent controllers in the future; today it is a very fast growing area.

2.1.2 From centralized to distributed automation

In industrial applications, informatics developed earlier when compared to private and science informatics. The mainframe computers were first used in economy control. In the next step, small controllers grew to powerful control and monitoring systems to control a single production line or a paper machine. In the third development line, programmable logic controllers (PLC) controlled a limited task at a complex production factory, which was then controlled and monitored by PC-level microcomputers. The fourth development line is now going on in the form of distributed automation: wireless sensor networks and intelligent actuators.

Because of security and latency problems, wireless technology is not yet in large use in industrial applications.

In any case, distribution gives another kind of security and reliability; if the structure of a distributed system is dynamic, it is allowing nodes to replace each other. The example of a bear and an ant nest shows this feature: when you are hunting with a gun in the forest, you can meet a bear and destroy this creature with a single accurate shot. You can also see an ant nest. With a single shot, you can make a slight disturbance in the life of the ant nest, but afterwards nobody can see any sign of your shot in the nest. The effect of one failure can destroy the whole system if it is centralized. In a distributed system, other resources replace the failed parts and it results in only a small disturbance in the normal operation.

2.1.3 From point-to-point to wireless mesh communication

Each one of the development lines described above has its own types of communication methods. The mainframe computers in universities had a lot of

(25)

wired ports to interface with teletype terminals. The only way to communicate was to use point-to-point wires.

In the second development line the most important way of communication was the internet realized by worldwide links, local area Ethernet and WLAN. The third development line in multimedia brought NMT, GSM, GPRS, 3G (UMTS, HSPA+), and 4G (LTE-A) technologies. Moreover, the 5G network is still under development for a new cellular network standard. One of the motivations for 5G is to support industrial applications by enhancing reliability and minimizing latency.

The standard development of 5G is under way and it has high objectives: the geographical coverage and availability are planned to be 100%. The planned connection speed is up to 10Gbps with 1 ms latency in end-to-end connection. One of the interesting features is the planned energy consumption: The lifetime for a battery powered machine-type device can be up to ten years. This is significant, when planning IoT-based sensor networks in future (Intelligence, 2014).

This second development line includes also multipoint factory buses to connect PLC’s and SCADA (Supervisory Control and Data Acquisition) devices. These kinds of systems are developed for industrial purposes.

One part of the fourth development line is the current technology in RFID and charge card applications. More smart applications are seeking their own form and standards for communication. The most common physical communication method is a short-range wireless link. The first network topologies are hierarchical, i.e. sensor networks. The ZigBee standard has some kind of mesh topology and TUTWSN (Tampere University of Technology Wireless Sensor Network) has full mesh topology (Kohvakka;Suhonen;Kuorilehto;Hännikäinen;& Hämäläinen, 2007). Perhaps the new ambient intelligence applications will use non-standard or very scalable standard wireless communication methods. These wireless technologies are seeking their own forms and features, depending on the applications they are to be used in.

2.1.4 From computing units to smart sensors

The structure in traditional automation can be divided into three parts: computing unit, interface and sensors/actuators. These are all separate and sometimes far from each other. The main intelligence and developed software was only in the computing unit. Along with the development of multifunctional microcontrollers, the intelligence is more and more in the sensors and actuators. In embedded ambient intelligence wireless devices, the whole system including sensors and even

(26)

actuators are on the same small circuit board sensing and acting according to acceleration, temperature, light and/or humidity.

The newest MEMS (Micro Electro Mechanical Sensor) technology makes it possible to integrate the sensors, actuators and user interface in a very small and intelligent form that can be used, for example, in a RF button. The MEMS chip is a silicon chip that has also mechanical structures. Force, rotation, pressure, weight, inclination and acceleration are very simple to measure. Actuators are not easy to realize in very small form; only micro motor and pump structures have been tested and they can be realized. Keyboards and command buttons can be replaced with acceleration, gyro and motion MEMS sensors. The user does not need to enter commands to a simple ambient intelligence device, but the behaviour of the user is the input needed for events. A text or graphical display is not needed in simple applications. Voice and light signs can replace them to give information on events happening in the ambience (Chang;Lee;& Chih-Yung, 2007).

For example, STmicroelectronics has different types of MEMS sensors:

acceleration, gyroscope, compass, microphone as well as environmental sensors:

pressure, temperature and humidity. The current applications for MEMS sensors are motion tracking, compass and navigation in smartphones and tablets. The gyro and acceleration sensors are used in gaming devices. Many fitness, wellness and home appliance applications use MEMS sensors. Crash detection and many smart sensors are used in car technology. Many robotic and medical applications also use MEMS sensors. Mostly the applications above use sensors to sense movement and position automatically similar to the user interface, but without the user interface.

(STMicroelectronics, 2016).

2.1.5 From computer connections to object connections

Mainframe computers are connected to each other and with terminals; this is required for the normal use of computers. This also means that generated scientific information is more widely spread.

The development of the current multimedia devices is focused on connecting people together with high communication speeds and versatile application software. The current trend seems to be new social media methods, virtualization and augmented reality. It means that people are connected also with a virtual or fictional world. (Palomäki, 2014b). In any case, this technology was left out of this thesis.

(27)

In ambient intelligence technology, the objects are connected wirelessly together.

There exists a huge set of applications to connect all kinds of objects with each other and with control systems. The objects are either moving or there are so many objects that wiring is not a worthwhile choice. It is a new idea to connect small numerous objects together in a smart way using wireless. This study focus on low- level connections where, as a minimum, an object can tell the wireless network at least: “I’m here” or “I see you”. The set of applications includes monitoring and the control of clothes, animals, children, personal estates, tools, toys, access and identification. Applications of this kind are quite new; excluding access control.

Connecting objects makes dumb things in our environment smart.

2.1.6 Smart dust

There are many technologies trying to distribute intelligence in smaller and smaller devices. One technology area is called ‘Smart dust’. The plan is to have wireless devices; which are tiny MEMS units with a set of sensors. The size of these motes can be the size of a grain of sand. They can process information and communicate with other neighbour motes up to a range of as much as 300m. The goal for researchers was to get chips with 1mm sides. The planned structure of a single smart dust mote is in Figure 3. (Hoffman, 2003)

Figure 3. Smart dust mote (Hoffman, 2003)

(28)

Many research groups are developing wireless Smart Dust technology around the world. Smart dust is actually a theoretical scenario in wireless technology, in which the intelligence is highly distributed. A single device would be an independent, intelligent measuring and communicating chip with the size of a millimetre cube.

The goal of these development projects is that these chips can be seeded in the ambience to sense conditions and to send the data forward wirelessly. There are also plans to develop some kinds of micro actuators for micro robotics. For example, a single smart dust chip can consist of a battery, a solar shell, a power capacitor, a controller with an analogy interface, sensors and optical components;

all the size of some cube millimetres. Optical communication is carried out with a MEMS laser diode and a CCR sensor. One example mote has been developed having the size of 63 mm3. (Warneke;Last;Liebowitz;& Pister, 2001)

The smallest independent working device is a RFID μ-Chip developed by Hitachi in Figure 4. The size of the chip is 0.4mm x 0.4mm. However, it is a RFID tag and needs an external magnetic field for power supply, and functions only as an identification chip. (Hitachi ltd, 2003)

Figure 4. RFID μ-Chip (Hitachi ltd, 2003)

Smart Dust gives the biggest benefit in space technology. Scientists are exploring a space telescope with swarms of particles. The swarms form floating lenses controlled by a laser, which are cheaper and lighter than conventional space telescopes. It is possible to form lenses even thousands of kilometres in diameter.

(Gawlowicz, 2014)

The other planned application for space technology is planet exploration. Particles, which are small enough, can fly in the wind. The computer chip controls the shape of a plastic sheath enabling it to steer the particles. Using wireless communication,

(29)

these particles form networks and include small sensors collecting information about the planet. (Rincon, 2007)

If smart dust scenarios are realized, the ambience can be intelligent. People and devices can have a view of the ambience via a smart dust wireless network and get important data on current or past conditions and events.

(30)

3 WIRELESS PROTOCOLS AND TECHNOLOGIES

3.1 Existing wireless technologies

Today’s wireless communication standards were developed to fulfil the existing demands of the time and anticipated future demands. However, all of the standards were developed some time ago, so it is uncertain, that they fit the present day demands or future ones. The existing technologies were studied to find out how suitable they are for distributed ambient intelligence. In this study, only wireless technology using free ISM (Industrial, Scientific and Medical) frequency bands was used. These bands can be used freely without license, but with restrictions on the allowed maximum effective transmission power (European Communications Office, 2016, ss. 115-116). The most used band is 2.4 GHz, therefore the focus was on technologies use this band.

3.1.1 Low Energy Bluetooth

L.M. Ericsson in Sweden started in 1994 the development of the first version of Bluetooth. In 1998 five companies formed a Special Interest Group, which grew in four years to a cooperation of 1500 companies. The basic aim was to connect computers wirelessly to peripheral devices. Bluetooth has been designed for high- speed point-to-point communication, but it has the same good features required for a wireless network. (Dursch;Yen;& Shih, 2004)

For small, battery powered control and monitoring applications; Bluetooth has a low-energy alternative BLE (Bluetooth Low Energy) standard developed by Bluetooth SIG (Special Interest Group). The structure of a Bluetooth network can be modified dynamically. One Bluetooth node senses the nearness of other nodes, adapts it to the network structure and starts the communication period. The network structure of Bluetooth is piconet, where one master connects with numerous slaves. Piconet is a basic network topology where multihopping is not possible. The classic Bluetooth enables routing between piconets (Scatternet), but in BLE it is no longer possible. A slave has a 32-bit identification address and theoretically, it can have a huge network size, but in practice in low-energy applications, the number of slaves is between 2 and 11. The bit rate in air is 1 Mbps, but the maximum application layer bit rate is 236.7 kbps. BLE requires memory resources in the controller: 40 kB ROM and 2.5 kB RAM. BLE technology is focused on being used, for example, in smartphones and in similar applications (Gomez;Oller;& Paradells, 2012).

(31)

BLE uses 40 RF channels in the ISM band 2.4 GHz, Transmission uses an adaptive frequency hopping mechanism between 37 available channels (Figure 5). Physical channels use a GFSK (Gaussian Frequency Shift Keying) modulation and the bit data rate in air is 1 Mbps. In wireless applications, where nodes are moving and the structure of the network is dynamic, the connection time of a single node is significant. The connection time (the latency delay) in the BLE network is usually from 7.5 ms up to 4000 ms (Gomez;Oller;& Paradells, 2012).

Figure 5. BLE channels (Nilsson, 2013)

The BLE standard was developed to give the sensors a lifetime from several months to several years using a single coin battery cell. The current consumption when transferring 16-64 bytes per second is about 40-90 uA (Kindt;Yunge;Diemer;& Chakraborty, 2014).

BLE is very suitable for simple sensor networks controlled, for example, by a laptop or smartphone. The possible applications are wearable technology and health monitoring devices. The lack of multi-hop limits the use of BLE in most ambient intelligent applications, because the whole network must be located inside the master’s range.

3.1.2 ZigBee

ZigBee is a wireless technology for industrial applications, especially in WSN (Wireless Sensor Network) applications. The ZigBee Alliance was established in 2001 the first ZigBee specification enabled more than 65000 nodes in a single

(32)

network. The standard IEEE 802.15.4 defines the physical layer and medium access control of ZigBee protocol architecture. The ZigBee protocol architecture stack includes the network layer and the application layer as shown in Figure 6 (Gomez & Paradells, 2010).

Physically, ZigBee has two node types: FFD (Full Function Device) and RFD (Reduced Function Device). FFD has normally a continuous power supply and includes all ZigBee features. RFD is low power, possibly a battery powered device having no routing features. The ZigBee network consists of one coordinator, several routers and end devices. A FFD can function as a coordinator or a network root, initializing the network and setting operational parameters as well as function as a router receiving and transmitting data between nodes. A RFD is normally an end device connected to one router or coordinator. Mostly the RFD is a battery powered sensor or a switch. In addition, a FFD can function as an end device with a continuous power supply and is more suitable as an actuator (Somani

& Patel, 2012).

Figure 6. ZibGee protocol architecture (Gomez & Paradells, 2010)

IEEE 802.15.4 standard was defined to use the 2.4 GHz band frequency area between 2405 MHz – 2480 MHz using sixteen 5 MHz channels (Figure 7). The modulation method is DSSS (Direct Sequence Spread Spectrum) using O-QPSK (Offset Quadrature Phase Shift Keying) (MaxStream, 2007).

(33)

Figure 7. IEEE 802.15.4 channels (MaxStream, 2007)

The ZigBee network is suitable for wireless sensor networks and other applications including sensors and actuators. Multimedia applications are not suitable for ZigBee, because its data rate is quite low. The most important advantages of ZigBee for ambient intelligence are the very low power consumption and the mesh topology of the network. The end nodes in the network can be battery-powered.

The data transmission range can be large in the mesh topology because of multihop routing. The size of a ZigBee network can be up to 216 (short address) or 264 (IEEE address) nodes, hence, in practice there are no limits to the number of supported nodes (Callaway, 2003).

ZigBee has been developed mainly for sensor networks, so there are some disadvantages concerning ambient intelligence applications. Only the RFD nodes are really low-power, because they have only one synchronized communication channel with one FFD node. The FFD nodes communicate both with their own RFD nodes and with other FFD nodes. The RFD node can be battery-powered, but the FFD normally has a continuous power supply. The ambient intelligence applications must take into account the power limits of these routing nodes.

A commercial ZigBee manufacturer must join the ZigBee Alliance and pay a membership fee of $4000 - $50000 per year to get the rights to use the ‘ZigBee’

trademark. The fees depend on, whether the member is an adopter, participant or promotor. This is a significant disadvantage for small and medium sized companies developing wireless technology (ZigBee Alliance, 2016)

(34)

3.1.3 6LowPAN

6LoWPAN (IPv6 over Low Power Wireless Personal Area Networks) is some kind of ZigBee extension. Protocol has been developed by the 6LowPAN Working Group to have access to ZigBee nodes via the Internet. This 6-byte Internet addressing mode gives interesting possibilities to control and monitor a personal area network. This technology gives a good way to realize IoT (Internet of Things) applications with ZigBee. 6LoWPAN is focused on ZigBee, but it is possible to take the same idea in use in other PAN area networks to realize a bridge between the Internet and the objects around us (Mulligan, 2007).

If 6LoWPAN network is larger and needs routing, it can function with mesh under routing or route over (Figure 8.). In “mesh under” routing, the routers are following IEEE 802.15.4 standard routing and are not using IP-addresses; the route is a single IP hop. In “route over”, every router has an IP address and so every hop is an IP-hop (Gomez & Paradells, 2010).

Figure 8. 6LoWPAN protocol architecture (Gomez & Paradells, 2010)

(35)

Z-Wave

Wireless Z-Wave technology is suitable for home control systems developed by the company ZenSys (a division of Sigma Design). The Z-Wave alliance controls the use of technology. A member can only be certified as a Z-Wave developer and the annual fees are 250$ to 4000$ depending on the membership level (Z-Wave alliance, 2017).

The Z-wave protocol uses different RF frequency bands: 868MHz (In Europe), 908MHz (In USA) and 2400MHz depending on the chip series. It uses BFSK modulation and the communication bit rates are 9.6kb/s, 40kb/s or 200kb/s (only with 2400MHz band), the range is from 30m (Indoor) to 100m (Outdoor). The Z- Wave protocol stack has 5 layers: physical, MAC, transfer, routing and application layers, as shown in Figure 9 (Gomez & Paradells, 2010).

Figure 9. Z-wave protocol architecture (Gomez & Paradells, 2010)

Figure 10. Z-Wave network example (Zensys A/S, 2006)

(36)

Z-Wave protocol consists of controlling devices and slave nodes. There are four different controller types: A portable controller, which can function, for example, as a remote controller in the system. A static controller, which can function, for example, as an internet gateway. An installer controller, which is required by the installer to set up the network. A bridge controller, which connects the Z-Wave to other different networks. There are three different Z-Wave slave types: The simplest slave cannot communicate directly with the other slaves but can receive commands, functioning for example as a light dimmer. A routing slave can send messages to other nodes and functions, for example, as a sensor. An enhanced slave, which has added resources compared with the routing slave, has clock and memory and can function, for example, as a weather station. The principal structure and node roles are shown in Figure 10. (Zensys A/S, 2006)

ANT protocol

Dynastream Innovations has developed the wireless communication protocol ANT for short range and very simple applications. ANT handles peer-to-peer, star, tree and fixed mesh topologies. It includes physical, network and transport layers of an OSI stack. In any ANT topology, there is one or more HUB nodes, which serves a host application controller. The network can include RELAY nodes as routers, which connect HUB nodes together. The sensor nodes can be in contact only with a fixed HUB node and they are primary transmitters i.e. masters of the communication channel. The lowest-power sensor has a one-way link and can only transmit, but sensors can also have a bidirectional link to the HUB. The communication channels are fully synchronized: the master can send within a defined time slot its data to the slave and the slave can response immediately if it uses a bidirectional channel (Dynastream Innovation Inc., 2014).

ANT protocol can use the frequency area 2400 MHz to 2524 MHz with 124 RF channels with 1 MHz increments. In an actual ANT chip there are only 78 RF channels using narrow band GFSK (Gaussian Frequency Shift Keying) modulation. The on-air bit rate is 1 Mbps and the maximum data byte rate between nodes is 20 kbps. It is possible to detect logical proximity between nodes using a maximum of 10 steps with the proximity search feature (Dynastream Innovation Inc., 2014) (Nordic Semiconductor, 2010a).

The ANT device can be in the form of a separate host controller (MCU) and ANT module. Some manufacturers have a SoC device (System on Chip), where the controller and ANT protocol stack are on the same chip, as shown in Figure 11 (Dynastream Innovation Inc., 2014).

(37)

The primary application of the ANT protocol is in the sport and fitness field, as indicated in Figure 12. It is very suitable for wearable technology to add intelligence to clothes and the body: to monitor health, condition and movement.

The ANT protocol was developed for a specific type of application. Therefore, it can include some limitations when using it in some ambient intelligence applications.

Figure 11. ANT layer alternatives (Dynastream Innovation Inc., 2014)

(38)

Figure 12. ANT application examples (ANT wireless, 2017)

3.1.4 TUTWSN

Tampere University of Technology has researched wireless sensor network technology in the DACI research group. This group has developed the wireless sensor network TUTWSN (Tampere University of Technology, 2009) with optimal features. The network topology is a full ad-hoc mesh. The number of nodes and multi-hops is unlimited. The network is fully distributed, i.e. it does not need any coordinator or external computing (Figure 13). As the network structure is modified, it repairs and reorganizes itself. TUTWSN has been developed especially to rule dynamic changes in a network, i.e. for mobile applications. This network is very low-powered, for example, the life-time of a badge node with 2 buttons and 3 LED lamps is 3-5 years using 2xAAA batteries.

(Kohvakka;Suhonen;Kuorilehto;Hännikäinen;& Hämäläinen, 2007) (Hämäläinen

& Hännikäinen, 2007)

Figure 13. TUTWSN topology (Hämäläinen & Hännikäinen, 2007)

TUTWSN is highly suitable for universal use in ambient intelligent applications.

There are more than 40 node types and also gateway nodes for Ethernet, Bluetooth and WLAN. The smallest node is a wrist node, which is suitable for a key chain.

(39)

The only disadvantage is that TUTWSN is a customized technology. It is not open source and today all applications are made by the TUTWSN research group.

3.1.5 nRF24 based technology

Nordic semiconductor wireless technology is suitable for customized applications.

Especially within hobbyist circles, the controllers nRF24LE1 and nRF24LU1 are widely used, for example, under the name Grazyradio. The older nRFL01 transceiver is still used, but it needs an extra controller. The top-rated application is Nano Quadcopter. (Bitcraze.io, 2016). These same controllers are used for wireless keyboards and mice.

nRF24 series technology uses the 2.4 GHz frequency area. The maximum bit rate in air is 2 Mbps. With 2 MHz wide channels, there are altogether 62 channels; 39 of them are inside the ISM band. With lower bit rates in air, 250 kbps or 1Mbps, there are 125 channels; 79 of them are inside the ISM band. The frequency hopping method is possible with user software. In addition, the controller includes an encryption/decryption accelerator to build safer protocols. RF communication uses narrow band GFSK (Gaussian Frequency Shift Keying) modulation, the same as in the ANT standard, but includes only the physical and data link layers of the OSI protocol stack. Using Enhanced ShockBurst with the auto retransmit feature, the user interface for the programmer is very simply and small applications were realized with quite light programming (Nordic Semiconductor, 2010b).

3.1.6 Other related technologies

3.1.6.1 NFC

Near Field Communication (NFC) is used for payments, using a credit card or mobile phone. The range is very short; mostly for security reasons. NFC technology, based on RFID technology, forms a ‘handshake’ between two devices that are near each other. RFID uses magnetic field induction in the 13.56MHz frequency band over a distance of up to 20 centimetres. NFC is used in passage control and product packaging. When NFC is installed in a mobile phone, it can read passive RFID tags; act as an active RFID tag or exchange data with other NFC devices. The NFC Forum is a non-profit industry association founded in 2004 to promote the standardization of NFC technology (NFC Forum 2007). (NFC Forum, 2016)

(40)

The NFC technology is partly suitable for ambient applications. Passive RFID tags have a very low-cost and are small enough to be implanted invisibly in clothes and the ambience. Mobile phones are very common and simple to use to function as a handheld module in a wireless network. However, it is not possible to create a sensor network using NFC technology.

The disadvantage of NFC is its very short range and the missing intelligence of RFID tags. The tags cannot communicate with each other and function independently, so the features of the applications are limited. In its original use, NFC is quite uncertain and misuse is possible. (Hakamäki & Palomäki, 2015)

3.1.6.2 UWB

Ultra Wideband (UWB) is a new interesting wireless technology. This wireless technology principle originates from 1942. The idea of the original UWB was to send very short data pulses as an unmodulated very wideband low-power signal.

This seems to be a good technology for a simple short range, battery powered wireless network. However, in reality it is not. From 2003 to 2006 a group of companies tried to define specifications for standard IEEE 802.15.3a. In 2005 some organizations merged into the WiMedia Alliance, which defined complete UWB specifications to develop solutions. The starting point of UWB seemed to be suitable for a simple, low-cost, low-power wireless network. However, UWB has been developed to replace high-speed Bluetooth and wired USB-connections between a PC and its peripherals. The result of UWB development is that it is no longer suitable for ambient intelligence applications (Aiello & Batra, 2006).

3.2 Comparison of wireless controllers

Electronic circuits give the framework for the features in a wireless network.

Comparing radio transceiver chips gives some information about how suitable they are for ambient intelligence applications. There are some main criteria for choosing the best possible hardware for applications. The layout of the electronics must be simple to fit into a small button or a wristwatch. In addition, the power source, i.e. the battery must be small. This means a highly integrated, single chip and ultra-low-power RF controller.

In this chapter, four different radio controllers are compared: Bluetooth, ZigBee, ANT and nRF24LE1. The compared Bluetooth radio controller comes from Atmel:

ATBTLC1000 (Atmel, 2016). The ZigBee controller comes from Texas Instruments: CC2630 (Texas Instruments, 2016). The other radio chips come from

(41)

Nordic Semiconductors nRF24AP2 transceiver (Nordic Semiconductor, 2010a) for the ANT sensor and the nRF24LE1 controller (Nordic Semiconductor, 2010b) for customized usage. All these use the 2.4 GHz ISM band. The basic comparison data is presented in Table 1. All transmit powers are defined at 0 dBm output. Transmit energy includes the transmission of a 32 bytes payload within a 64 bit frame using 3V power voltage.

(42)

Table 1. Comparison between RF controllers

Technology Bluetooth ZigBee ANT nRF

Controller ATBTLC1000 CC2630- RSM

nRF24AP2 nRF24LE1

Pin count 32 32 32 24

IO-pins 14 10 7 7 AD (bits) 2 (11 bit) 5 (12 bit) 7 (12 bit) 7 (12 bit)

Size / mm 4 x 4 4 x 4 5 x 5 4 x 4

Power down 0.95 μA 0.15 μA 0.5 μA 0.5 μA

Sleep power 1.35 uA 1,0 uA 3.0 uA 1.0 uA Active power 0.85 mA 1.95 mA 2.5 mA 2.5 mA Transmit power 3.0 mA 6.1 mA 15 mA 11.1 mA Receive power 4.0 mA 5.9 mA 17 mA 13.3 mA Max bit rate 1 Mbps 250kbps 1 Mbps 2 Mbps Processor core Cortex M0 Cortex M3 external 8051+

Transmit energy 2.9 μJ 23.4 μJ 14.4 μJ 6.4 μJ

RSSI yes yes no no

There are some differences between the chips. The pin count tells something about the complexity of the functions, the demands of the electronics assembly work and the price. A current near 1 μA in the power-down state is not significant, because the self-discharge of batteries is sometimes larger. In the Bluetooth and ZigBee controllers, the processor core is very powerful in comparison with the nRF24LE1 and it is needed to control the complete wireless standard. The ANT transceiver needs an external controller, which takes extra space in small layouts. For the research work described in this thesis, the nRF24LE1 radio controller is used, because the ambient intelligence applications need highly customized features to test separate protocols without limits and the laboratories at Seinäjoki University of Applied Sciences have already a programming environment developed for this controller.

3.3 Internet of Things

The Internet of Things (IoT) is the network of objects exchanging data with each other. It allows objects to be monitored and controlled remotely via an infrastructure network. Every object has an individual IP address to be controlled via the Internet. The IDC organization estimated that the market for IoT solutions

(43)

would reach 7.1 trillion USD in 2020. The biggest part of IoT is M2M (Machine to Machine) communication. D2D (Device to Device) communication defines peer- to-peer communication between devices which belongs also to the IoT technology (Kalyani & Sharma, 2015).

3.3.1 M2M

The term M2M is widely used by industry. It defines all communication between machines using either wired or wireless methods. If M2M is in a wireless form, it uses low power wide-area network connections (LPWAN) to route or to connect machines together. These kinds of protocols are, for example, BLE, ZigBee or Wi- Fi. The IDC organization estimated that the LPWAN connections would grow to 5 billion by 2022 (Kalyani & Sharma, 2015).

M2M communication is very suitable for the industry. All kinds of monitoring, controlling, servicing, firmware updating and software developing is possible remotely via the Internet. It also opens the world-wide market for smaller companies. It gives great possibilities to centralize different types of know-how and combine different specialists from every corner of the world around one single product.

3.3.2 D2D

The idea of IoT and M2M is that the structure of the network is hierarchical: A higher level in the network can control lower level objects and collect data from them via routers. The device to device (D2D) communication includes also peer- to-peer communication. It means that the devices are not dependant on the network structure, but can exchange data directly with each other without routing.

As an example, a cellular connection needs a network structure to work inside a network range, but using D2D, the direct connection between phones is possible without any cellular network (Kalyani & Sharma, 2015).

D2D communication is suitable for mobile objects when the connection to the network is working randomly or is changeable. The wireless range can be much more limited, than in M2M, therefore the devices can be very low-powered and small-sized to be better embedded in objects. The objects are not necessarily visible on the Internet. This kind of solution is, for example, traffic vehicles: as two cars pass each other, they can exchange weather information or problems on the route. RF buttons on wild animals can ‘gossip’ information about travel routes via other animals to a collecting device that is located at the feeding grounds and can

Wireless distributed intelligence in personal applications